English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Event cameras capture visual information with a high temporal resolution anda wide dynamic range. This enables capturing visual information at fine timegranularities (e.g., microseconds) in rapidly changing environments. This makesevent cameras highly useful for high-speed robotics tasks involving rapidmotion, such as high-speed perception, object tracking, and control. However,convolutional neural network inference on event camera streams cannot currentlyperform real-time inference at the high speeds at which event cameras operate -current CNN inference times are typically closer in order of magnitude to theframe rates of regular frame-based cameras. Real-time inference at event camerarates is necessary to fully leverage the high frequency and high temporalresolution that event cameras offer. This paper presents EvConv, a new approachto enable fast inference on CNNs for inputs from event cameras. We observe thatconsecutive inputs to the CNN from an event camera have only small differencesbetween them. Thus, we propose to perform inference on the difference betweenconsecutive input tensors, or the increment. This enables a significantreduction in the number of floating-point operations required (and thus theinference latency) because increments are very sparse. We design EvConv toleverage the irregular sparsity in increments from event cameras and to retainthe sparsity of these increments across all layers of the network. Wedemonstrate a reduction in the number of floating operations required in theforward pass by up to 98%. We also demonstrate a speedup of up to 1.6X forinference using CNNs for tasks such as depth estimation, object recognition,and optical flow estimation, with almost no loss in accuracy.

EvConv: Fast CNN Inference on Event Camera Inputs For High-Speed Robot Perception